By role · Content Creator

The Content Creator's Claude Stack: From Monthly Calendar to Published Reel

Published April 29, 2026 · 9 min read

The content creator's actual problem has never been ideas. The problem is the loop. You sit down on Sunday, brainstorm thirty topics, write hooks for ten of them, draft captions for five, pick three to film, then run out of time before you've even thought about repurposing. By Wednesday the calendar is dust again. Saturday morning you're filming whatever you can remember instead of whatever you planned.

Most "AI for creators" advice papers over this with a single magic prompt that promises to write a viral hook. That's not the unlock. The unlock is wiring six prompts into a chain so the calendar feeds the hooks, the hooks feed the scripts, the scripts feed the captions, and the captions feed the repurposing machine — all sharing one voice profile so it sounds like you across every platform. Once you've got that wired, four hours on Monday produces enough content to last the rest of the week without thinking.

This is the stack. Six prompts, in order, with the exact wiring between them.

The setup that makes the whole thing work

Before any of these prompts, you need a Claude Project (or one master system prompt if you're on the API). It should contain three things: your audience profile, your two voice examples, and your three pillars. That's it. No long instructions. No persona-heavy roleplay. Just the source material every other prompt will inherit from.

AUDIENCE
[two sentences about who watches/reads/follows you]

VOICE — match the energy of these two pieces:
EXAMPLE 1: [paste your best-performing post]
EXAMPLE 2: [paste another that landed, different topic]

PILLARS
1. [topic 1]
2. [topic 2]
3. [topic 3]

Whenever I ask you to write something, default to this
voice and these pillars unless I tell you otherwise.

This is the foundation. Every other prompt in the stack assumes this Project is already loaded. If you skip this step, every output you get back will sound like generic creator advice and you'll spend twenty minutes editing it into your voice. With this loaded, edits drop to two minutes per piece because Claude already knows the rules.

Prompt 1 — The monthly content calendar

You run this once a month. It produces 30 days of post ideas mapped to your pillars, with a 2-3 sentence concept for each, sized for the format you're shipping on (reel, carousel, post, newsletter, whatever). Run on the 1st of every month. Keep the output in a doc.

Build me a 30-day content calendar for [Instagram /
TikTok / YouTube Shorts / X / newsletter].

Spread the topics across my three pillars roughly
evenly. Mix the angles — some how-to, some opinion,
some story, some behind-the-scenes, some debate-bait,
some "saved post" educational. Don't repeat angle
types more than 2 days in a row.

For each day output:
- Day number
- Pillar tag
- Format (reel / carousel / single post)
- Working hook (one sentence, not a title)
- Concept (2-3 sentences)

Constraint: assume I'll use the hook generator and
script writer downstream. Concepts should be
specific enough to script from, but not so detailed
that I lose flexibility.

The output is your month's skeleton. You'll change ~20% of it as the month progresses (real life happens, news breaks, you get tired of a topic). That's fine. The other 80% saves you from staring at a blank doc on Monday morning trying to remember what you wanted to talk about.

Prompt 2 — The hook generator

Once you've got the day's concept, you run this to get ten hook variations. Run it the morning of the day you're filming, not on the calendar day, because the right hook depends on what the algorithm is rewarding that week.

Give me 10 hook variations for this concept:

[paste concept from calendar]

Mix:
- 3 curiosity-gap hooks ("the thing nobody tells you...")
- 2 contrarian hooks ("everyone says X. I think X is
  wrong.")
- 2 specific-number hooks ("I tested this for 30 days")
- 2 question hooks (real questions, not engagement-bait)
- 1 story hook (one specific moment as the hook)

Each hook should be under 12 words and read like
something a real person would say out loud, not like
a YouTube title. No "you won't believe," no "shocking,"
no all-caps. Match my voice from the examples.

You pick one or two and ship them. The rest go in a "future hooks" doc — they'll seed next month's calendar. Hooks compound across months even if you don't use them today.

Prompt 3 — The reel script writer

This is the prompt that turns the hook into 25-45 seconds of speakable script. It should sound like you, not like a teleprompter.

Write me a reel script for this concept and hook.

Concept: [paste]
Hook: [paste the chosen hook from prompt 2]

Format:
- Hook line (the one you wrote — keep verbatim)
- 3-4 beats of payoff, each 8-15 words
- One specific example or number in the middle
- A close that either teases a follow-up post or
  asks one direct question

Constraint: total reads in 30 seconds when said out
loud at normal pace. No filler ("now," "so basically,"
"the thing is"). No "smash that follow." Read it back
in your head before sending — does it sound like
something I'd actually say?

The "read it back" instruction matters. It's a tiny self-check that nudges Claude to revise its own output before you see it. The drafts that come out of this prompt usually need one or two word swaps, not a rewrite.

Prompt 4 — The caption + hashtag pack

You've got the script. Now you need the caption that runs underneath. The mistake almost every creator makes is using the script as the caption. Don't. The reel does the visual work. The caption does the saved-post work — depth, context, the part viewers screenshot.

Here's the reel script:
[paste]

Write the caption that goes under it.

Format:
- One opening line that re-states the hook with a
  small twist (so it works as a hook again for
  people reading first)
- 3-5 short paragraphs that expand on the script with
  one extra example or detail the reel didn't have time
  for
- A close that's either a question (real, not
  engagement-bait) or a soft CTA (newsletter / link
  in bio / save for later)

Then list 12 hashtags: 4 broad (1M+ posts), 6 mid
(50k-300k), 2 niche (under 10k). All actually
relevant to my pillars.

Constraint: caption reads on its own without the
reel. Someone who never watched the video should
still walk away with one specific thing. Match my
voice.

This is what makes a reel save-able instead of just watch-able. The save rate is what tells the algorithm to push you. Treat the caption as primary content, not filler.

Prompt 5 — The carousel slide outliner

Some of your best calendar entries deserve a carousel instead of (or in addition to) a reel. Carousels are the highest-saved format on most platforms in 2026 because people screenshot frames as references. This prompt turns a concept into a 7-slide outline.

Turn this concept into a 7-slide carousel outline.

Concept: [paste]

Slide 1: cover — single sharp claim or number, no
nuance, no setup
Slide 2: the problem (what most people get wrong)
Slide 3-5: three steps / examples / lessons (one per
slide, each with one specific number or named thing)
Slide 6: the rule / takeaway (one sentence, screenshot-
worthy)
Slide 7: soft CTA (save / share / follow for more on
this pillar)

For each slide give me:
- Headline (max 7 words, fits on one line at 60pt)
- Body (max 25 words, fits on one line at 24pt)
- Image cue (1 sentence — what should be on screen)

Constraint: any single slide should make sense alone
if reposted. Match my voice.

The "any single slide makes sense alone" constraint is the secret. It's how you get reposted — people screenshot one slide, and that one slide doubles as a complete idea. If your slide 4 only makes sense after slide 3, you've left engagement on the table.

Prompt 6 — The repurpose machine

This is the one creators almost never set up, and it's the one that quietly produces the most output for the least work. You take one piece of long-form content (a video, a podcast, a newsletter) and break it into four short pieces for different platforms.

Here's a [video transcript / newsletter / podcast
chapter] I just published:

[paste]

Generate 4 repurposed pieces from it:

1. A 30-second reel script (hook + 3 beats + close)
   pulling out the most quotable moment.
2. A 5-tweet thread with one stat or specific moment
   per tweet, ending in a soft CTA back to the original.
3. A LinkedIn post (200-300 words) angled for an
   adjacent professional audience — strip the casual
   energy a notch, add one piece of business framing.
4. A 7-slide carousel outline using the format from
   prompt 5.

Constraint: each piece must work standalone without
referencing the original ("as I said in my video..."
is banned). Each piece should make someone want to
go back and watch / read the source.

Match my voice across all 4. Drop the LinkedIn voice
to about 70% energy on the LinkedIn one only — slightly
less casual but still recognizably me.

That last instruction matters. LinkedIn punishes IG/TikTok energy — too casual reads as unprofessional in that feed — but most creators overcorrect and end up with bland LinkedIn voice. The "70% energy" framing keeps your voice intact while adjusting the dial just enough.

Run this prompt every time you ship a long-form piece. You'll get four extra distribution surfaces from work you've already done. The math is brutal — one hour of long-form work plus this prompt produces what most creators struggle to produce in a full week of original short-form.

The time math (why this works)

Before this stack, the average solo creator spends roughly 18 hours a week on content: 4 hours brainstorming, 4 hours writing scripts/captions, 4 hours filming/recording, 4 hours editing, 2 hours repurposing. With the stack wired through a Claude Project:

• Calendar (Prompt 1): 20 minutes a month, 5 minutes per week amortized
• Hooks (Prompt 2): 5 minutes per shoot day
• Script (Prompt 3): 5 minutes per piece
• Caption (Prompt 4): 5 minutes per piece
• Carousel (Prompt 5): 10 minutes per piece, used 1-2x a week
• Repurpose (Prompt 6): 10 minutes per long-form piece

Net: about 4 hours a week on the writing side. The filming and editing time stays the same — Claude doesn't film for you. But you've cut the cognitive load that used to spread filming days into entire weekends. That's the real unlock — not faster output, but a smaller mental footprint.

The four ways this breaks (and how to fix each)

Voice drift on long sessions

Claude can drift toward generic creator-advice voice if your conversation runs over 20 messages. When that happens, paste your two voice examples again with one line: "reset to this voice." It snaps back. Don't try to fight drift with adjectives — go back to the source material.

Hooks that all sound the same

If Prompt 2's outputs start feeling samey, it's because Claude has locked onto a single hook template from your examples. Solution: in your Project, add a third voice example that's structurally different (a story-format post if your other two are how-tos, or vice versa). The model needs at least one outlier to know which dimensions to vary.

Carousel slides that read like a listicle

If your carousel outputs feel listy and forgettable, it's because the prompt is generating headlines without conflict. Add to Prompt 5: "Each slide should contain one tension — something the reader didn't expect or didn't agree with." The carousels start carrying weight again.

Repurposing that flattens your voice for LinkedIn

If the LinkedIn version of Prompt 6 keeps coming back too corporate, the "70%" framing isn't enough. Add: "Keep the contractions. Keep one self-deprecating beat. No 'I'm thrilled,' no 'humbled,' no 'amazing team.'" That short ban-list does more than any positive instruction.

What to do this week

You don't need to wire the whole stack at once. Start with the foundation Project and Prompts 1, 3, and 4. That's the calendar, the script, and the caption — the three you'll use every shoot day. Get those running smoothly for two weeks. Then add Prompt 6 (repurposing) because it's the highest-leverage addition. Save Prompts 2 and 5 for once the rest is muscle memory.

The whole stack starts paying for itself the moment you stop staring at a blank doc on Monday morning. That's usually within the first week.

The creators who make this look easy aren't faster typers and they don't have better ideas. They've wired the loop so the loop runs them. This is the wiring.